erv 0.0.2 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 0116fa6cb4d3e6b11d3602ed2608b6b60a90f717
-  data.tar.gz: aac7790d8d9f9ab82e5ed0c70157cc7e95ef3398
+  metadata.gz: 4df7f1fe74e08a93b0011dbce05db2862db6f742
+  data.tar.gz: bef0cb26a91b1353971c3d2f1de8cc071832b60b
 SHA512:
-  metadata.gz: 94cc7f3280ae54020481e86d0e1caa181b539b0d8d6ca31e436e0eb4ed6ff0c12d000b9fc8443e32cb137aec706ef3b2831b1676d1751959c1c333cf72035ab4
-  data.tar.gz: 5fb06905713c33f6fd563849c7c44ceae643a272cc52250a4a41cdaa6c01a9a79e334abe1d679a13ce833820e4b58b045f68d3b619d8fd123e6fbbd806a01904
+  metadata.gz: 85bc816dcaf649c8410138ddcb3496d449c701e9f847d842139c59c8e17a5f0e97b857ca362845c7365138afe0b524902bbdfd1eb060aa4151b03246582b1298
+  data.tar.gz: 4308c0e5862acb574b6adb771998e95459f19dc929749716f55118ac068246f319be8bcd3a801ee36d6aaa2a812a6c94ed08f41efcb7568773624c4a0a1c95b5

data/README.md

@@ -10,31 +10,13 @@ variables with a given probability distribution (gaussian, uniform, etc.) and
 to sample from them. ruby-erv was built from code that I extracted out of
 several scientific software I wrote for my research projects.
 
-ruby-erv is designed to work on both YARV/MRI and JRuby. I have not tested it
-on Rubinius yet.
-
 
 ## Installation
 
-I have not released ruby-erv on RubyGems, yet. For the moment, if you want to
-try it just place this line:
-
-```ruby
-gem 'erv', git: 'https://github.com/mtortonesi/ruby-erv.git'
-```
-
-in your Gemfile and run:
-
-    bundle install
-
-If using JRuby, you'll also need to run:
-
-    rake get_latest_commons_math_snapshot
+You can get the stable version of ruby-erv by installing the erv gem from
+RubyGems:
 
-to fetch the latest version of Apache Commons Math 3.3-SNAPSHOT. See the
-[implementation notes below](#implementation-notes) for more information. (Note
-that the rake-based installation of Apache Commons Math 3.3-SNAPSHOT requires
-[nokogiri](http://nokogiri.org/).)
+    gem install erv
 
 
 ## Examples
@@ -45,49 +27,39 @@ variables using ruby-erv, and how to sample from them:
 ```ruby
 require 'erv'
 
+# Gaussian random variable with mean 10 and standard deviation 2
 gaussian_rv = ERV::RandomVariable.new(distribution: :gaussian,
                                       mean: 10, sd: 2)
 s1 = gaussian_rv.sample
+
+# Geometric random variable with probability of success 0.3
+geometric_rv = ERV::RandomVariable.new(distribution: :geometric,
+                                       probability_of_success: 0.3)
+s2 = geometric_rv.sample
+```
+
+Starting from version 0.2.0, ruby-erv also supports mixture models. Here is a
+simple example of how to create a random variable with an exponential mixture
+distribution:
+
+```ruby
+require 'erv'
+
+emd = ERV::MixtureDistribution.new([ { distribution: :exponential, rate: 1.0, weight: 100.0 },
+                                     { distribution: :exponential, rate: 2.0, weight: 200.0 },
+                                     { distribution: :exponential, rate: 3.0, weight: 300.0 } ])
+s3 = emd.sample
 ```
 
 
 ## Implementation notes
 
-In YARV/MRI, ruby-erv leverages the GNU Scientific Library (GSL) for random
-number generation according to the desired probability distribution.
-Unfortunately, ruby-gsl is unmaintained and broken. (See [the
-patch](http://rubyforge.org/tracker/?func=detail&atid=1169&aid=29353&group_id=285)
-that I sent them ages ago and that was never merged.) So, ruby-erv uses the
-[gsl-nmatrix](https://github.com/SciRuby/rb-gsl) fork of ruby-gsl.
-
-In JRuby, ruby-erv leverages the [Apache Commons
-Math](http://commons.apache.org/proper/commons-math/) library to access the
-same random number generation functions. Unfortunately, at the time of this
-writing the Apache Commons Math maintainers have not shipped the 3.3 release
-yet. Commons Math 3.3 should include support for geometric distribution (that I
-need for several of my projects), thanks to [a
-patch](https://issues.apache.org/jira/browse/MATH-973) that I submitted and
-that was merged some time ago. So, for the moment ruby-env builds on top of the
-3.3-SNAPSHOT version of Apache Commons Math.
-
-To facilitate the installation of ruby-env under JRuby, I have decided to
-bundle the jar archive of Apache Commons Math 3.3-SNAPSHOT in (the jars
-directory of) the ruby-env gem package. This is a rather dirty but not uncommon
-approach, as many other gems (including the awesome
-[Nokogiri](https://github.com/sparklemotion/nokogiri/tree/master/lib)) bundle
-external jar dependencies in their JRuby version. However, in future I might
-decide to switch to a more powerful, Maven-based automated installation of
-Apache Commons Math 3.3-SNAPSHOT. For instance, based on Christian Meier's
-[jar-dependencies](https://github.com/mkristian/jar-dependencies).
-
-
-## Acknowledgment
-
-I would like to thank [Christian Meier](https://github.com/mkristian) for his
-very valuable suggestions on how to package this gem for JRuby. If you're
-interested in building serious applications for JRuby, I strongly recommend you
-to check out Christian's [jbundler](https://github.com/mkristian/jbundler) and
-[jar-dependencies](https://github.com/mkristian/jar-dependencies).
+Starting from version 0.2.0, ruby-erv makes use of the standard (Pseudo) Random
+Number Generator provided by Ruby VMs, which is a variant of the
+Mersenne-Twister algorithm modified to have a period of 2<sup>19937</sup>-1.
+The randomness provided by Ruby's PRNG, whose adoption significantly improved
+ruby-erv's portability to different Ruby VMs (MRI, JRuby, etc.), should be more
+than enough for most purposes.
 
 
 ## License

data/Rakefile

@@ -1,55 +1,11 @@
-if RUBY_PLATFORM =~ /java/
+require 'bundler/gem_tasks'
 
-  require 'lib/erv/version'
-  require 'bundler/gem_helper'
-
-  require 'nokogiri'
-  require 'open-uri'
-
-  # setup absolute path for jars directory
-  JAR_DIR = File.expand_path(File.join(File.dirname(__FILE__), 'jars'))
-
-  directory JAR_DIR
-
-  desc 'Get latest snapshot of Apache Commons Math 3.3'
-  task :get_latest_commons_math_snapshot => [ JAR_DIR ] do
-    # base URL of the apache commons math 3 snapshot repository
-    BASE_URL = 'http://repository.apache.org/content/groups/snapshots/org/apache/commons/commons-math3/3.3-SNAPSHOT/'
-
-    # retrieve timestamp and build number of latest snapshot
-    puts 'Retrieving metadata information on latest Apache Commons Math 3.3 snapshot.'
-    doc = Nokogiri::HTML(open(BASE_URL + 'maven-metadata.xml'))
-    timestamp = doc.xpath('//timestamp/text()')
-    buildnumber = doc.xpath('//buildnumber/text()')
-
-    # get archive name of latest snapshot
-    filename = "commons-math3-3.3-#{timestamp}-#{buildnumber}.jar"
-
-    # get destination file name
-    jar_path = File.join(JAR_DIR, filename)
-
-    # don't download if file already exists
-    if File.exists?(jar_path)
-      puts 'Archive already present in jars directory.'
-    else
-      puts 'Retrieving latest Apache Commons Math 3.3 snapshot.'
-      file_url = BASE_URL + filename
-      File.write(jar_path, open(file_url).read)
-    end
-
-    # clean JAR_DIR of every file except filename
-    puts 'Removing obsolete archives from jars directory.'
-    FileUtils.rm(Dir.glob("#{JAR_DIR}/*").reject!{|file| file == jar_path })
-  end
-
-  desc "Build erv-#{ERV::VERSION}-java.gem into the pkg directory."
-  task :build => [ :get_latest_commons_math_snapshot ] do
-    Bundler::GemHelper.instance = Bundler::GemHelper.new
-    Bundler::GemHelper.instance.build_gem
-  end
-
-else # RUBY_PLATFORM !~ /java/
-
-  require 'bundler/gem_tasks'
+require 'rake/testtask'
 
+Rake::TestTask.new do |t|
+  t.libs << 'test'
+  t.test_files = Dir.glob('test/**/*_test.rb').sort
+  t.verbose = true
 end
+
+task :default => :test

data/TODO

@@ -0,0 +1,2 @@
+* add specs
+* documentation

data/erv.gemspec

@@ -18,16 +18,9 @@ Gem::Specification.new do |spec|
   spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
   spec.require_paths = ['lib']
 
-  spec.add_development_dependency 'bundler', '~> 1.3'
-  spec.add_development_dependency 'rake'
-  spec.add_development_dependency 'rspec'
-
-  if RUBY_PLATFORM =~ /java/
-    # we need to release a JRuby-specific gem
-    spec.platform = 'java'
-    # include Apache Commons Math 3.3 jar archive
-    spec.files.concat(Dir['jars/*.jar'])
-  else
-    spec.add_dependency 'gsl-nmatrix'
-  end
+  spec.add_development_dependency 'bundler', '~> 1.14'
+  spec.add_development_dependency 'rake', '~> 12.0'
+  spec.add_development_dependency 'minitest', '~> 5.10'
+  spec.add_development_dependency 'minitest-reporters', '~> 1.1'
+  spec.add_development_dependency 'minitest-spec-context', '~> 0.0.3'
 end

data/lib/erv/constant_distribution.rb

@@ -7,8 +7,19 @@ module ERV
       super(opts)
 
       raise ArgumentError unless opts[:value]
-      val = opts[:value].to_f
-      @func = Proc.new { val }
+      @val = opts[:value].to_f
+    end
+
+    def mean
+      @val
+    end
+
+    def variance
+      0.0
+    end
+
+    def sample
+      @val
     end
   end
 

data/lib/erv/distribution.rb

@@ -1,15 +1,15 @@
-require 'erv/rng'
-
 module ERV
 
   class Distribution
     def initialize(opts)
-      # create RNG object
-      @rng = RNG.make_rng(opts[:seed])
-    end
-
-    def sample
-      @func.call
+      # use provided RNG or create a new one
+      if opts[:rng]
+        @rng = opts[:rng]
+      elsif opts[:seed]
+        @rng = Random.new(opts[:seed])
+      else
+        @rng = Random.new
+      end
     end
   end
 

data/lib/erv/exponential_distribution.rb

@@ -1,30 +1,28 @@
-if RUBY_PLATFORM == 'java'
-  require 'java'
-  JExponentialDistribution = org.apache.commons.math3.distribution.ExponentialDistribution
-end
-
 require 'erv/distribution'
+require 'erv/support/try'
 
 
 module ERV
 
   class ExponentialDistribution < Distribution
+    attr_reader :mean, :variance
+
     def initialize(opts)
       super(opts)
 
-      raise ArgumentError unless opts[:mean]
-      mean = opts[:mean].to_f
-
-      if RUBY_PLATFORM == 'java'
-        # create distribution
-        d = JExponentialDistribution.new(@rng, mean,
-                JExponentialDistribution::DEFAULT_INVERSE_ABSOLUTE_ACCURACY)
-        # setup sampling function
-        @func = Proc.new { d.sample }
-      else
-        # setup sampling function
-        @func = Proc.new { @rng.exponential(mean) }
-      end
+      @rate = opts[:rate].try(:to_f)
+      raise ArgumentError unless @rate and @rate > 0.0
+
+      @mean = 1 / @rate
+      @variance = @mean ** 2
+    end
+
+    def sample
+      # starting from a random variable X ~ U(0,1), which is provided by the
+      # RNG, we can obtain a random variable Y ~ Exp(\lambda), with mean = 1 /
+      # \lambda, through the transformation: Y = - (1 / \lambda) ln X. see
+      # [GROESE11], section 4.2.3.
+      - @mean * Math.log(@rng.rand)
     end
   end
 

data/lib/erv/gaussian_distribution.rb

@@ -1,8 +1,3 @@
-if RUBY_PLATFORM == 'java'
-  require 'java'
-  java_import org.apache.commons.math3.distribution.NormalDistribution
-end
-
 require 'erv/distribution'
 
 
@@ -13,19 +8,28 @@ module ERV
       super(opts)
 
       raise ArgumentError unless opts[:mean] and opts[:sd]
-      mean = opts[:mean].to_f
-      sd   = opts[:sd].to_f
-
-      if RUBY_PLATFORM == 'java'
-        # create distribution
-        d = NormalDistribution.new(@rng, mean, sd,
-                NormalDistribution::DEFAULT_INVERSE_ABSOLUTE_ACCURACY)
-        # setup sampling function
-        @func = Proc.new { d.sample }
-      else
-        # setup sampling function
-        @func = Proc.new { mean + @rng.ran_gaussian(sd) }
-      end
+      @mu    = opts[:mean].to_f
+      @sigma = opts[:sd].to_f
+    end
+
+    def sample
+      # use box-muller algorithm (see [GROESE11], section 4.2.11, algorithm
+      # 4.47) to obtain x ~ N(0,1).
+      u_1 = @rng.rand
+      u_2 = @rng.rand
+      x = Math.sqrt(-2.0 * Math.log(u_1)) * Math.cos(2.0 * Math.PI * u_2)
+
+      # use location-scale transformation to obtain a N(\mu, \sigma^2)
+      # distribution. see [GROESE11], section 3.1.2.2.
+      @mu + @sigma * x
+    end
+
+    def mean
+      @mu
+    end
+
+    def variance
+      @sigma ** 2
     end
   end
 

data/lib/erv/mixture_distribution.rb

@@ -0,0 +1,81 @@
+require 'erv/constant_distribution'
+require 'erv/discrete_uniform_distribution'
+require 'erv/exponential_distribution'
+require 'erv/gamma_distribution'
+require 'erv/gaussian_distribution'
+require 'erv/general_pareto_distribution'
+require 'erv/geometric_distribution'
+require 'erv/uniform_distribution'
+
+module ERV
+
+  class MixtureDistribution
+    def initialize(confs, opts={})
+      raise ArgumentError, "Please, provide at least 2 distributions!" unless confs.length >= 2
+
+      @mixture = []
+      @weight_sum = 0.0
+      while dist_conf = confs.shift
+        # get weight ...
+        weight = dist_conf.delete(:weight).to_f
+
+        # ... and keep track of it
+        @weight_sum += weight
+
+        # get distribution name
+        dist_name = dist_conf.delete(:distribution).to_s
+
+        # get class name that corresponds to the requested distribution
+        klass_name = dist_name.split('_').push('distribution').map(&:capitalize).join
+
+        # create distribution object
+        distribution = ERV.const_get(klass_name).new(dist_conf)
+
+        # add distribution to mixture
+        @mixture << { weight: weight, distribution: distribution }
+      end
+
+      seed = opts[:seed]
+      @mixture_sampler = (seed ? Random.new(seed) : Random.new)
+    end
+
+    def sample
+      x = @mixture_sampler.rand
+
+      # find index of distribution we are supposed to sample from
+      i = 0
+      while x > (@mixture[i][:weight] / @weight_sum)
+        x -= (@mixture[i][:weight] / @weight_sum)
+        i += 1
+      end
+
+      # return sample
+      @mixture[i][:distribution].sample
+    end
+
+    def mean
+      @mean ||= calculate_mean
+    end
+
+    def variance
+      @variance ||= calculate_variance
+    end
+
+    private
+
+      def calculate_mean
+        @mixture.inject(0.0) do |s,x| 
+          s += (x[:weight] / @weight_sum * x[:distribution].mean)
+        end
+      end
+
+      def calculate_variance
+        @mixture.inject(0.0) do |s,x| 
+          s += (x[:weight] / @weight_sum * 
+                ((x[:distribution].mean - self.mean) ** 2 + 
+                 x[:distribution].variance))
+        end
+      end
+  end
+
+end

data/lib/erv/random_variable.rb

@@ -18,6 +18,7 @@ module ERV
 
       # get class name that corresponds to the requested distribution
       klass_name = dist_name.split('_').push('distribution').map(&:capitalize).join
+
       # create distribution object
       @dist = ERV.const_get(klass_name).new(opts)
     end
@@ -30,9 +31,9 @@ module ERV
 
   class SequentialRandomVariable
     def initialize(args)
-      @first = args.delete(:first_value)
-      @most_recent = @first || 0.0
-      @var = RandomVariable.new(args)
+      first = args[:first_value]
+      @most_recent = first.nil? ? 0.0 : first
+      @var = RandomVariable.new(args.reject{|k,v| k == :first_value })
     end
 
     def next

data/lib/erv/uniform_distribution.rb

@@ -1,8 +1,3 @@
-if RUBY_PLATFORM == 'java'
-  require 'java'
-  java_import org.apache.commons.math3.distribution.UniformRealDistribution
-end
-
 require 'erv/distribution'
 require 'erv/support/try'
 
@@ -15,18 +10,15 @@ module ERV
 
       raise ArgumentError unless opts[:max_value]
       max = opts[:max_value].to_f
-      min = opts[:min_value].try(:to_f) || 0.0
+      @min = opts[:min_value].try(:to_f) || 0.0
+      @range = max - @min
+    end
 
-      if RUBY_PLATFORM == 'java'
-        # create distribution
-        d = UniformRealDistribution.new(@rng, min, max,
-                UniformRealDistribution::DEFAULT_INVERSE_ABSOLUTE_ACCURACY)
-        # setup sampling function
-        @func = Proc.new { d.sample }
-      else
-        # setup sampling function
-        @func = Proc.new { @rng.flat(min, max) }
-      end
+    def sample
+      # starting from a random variable X ~ U(0,1), which is provided by the
+      # RNG, we can obtain a random variable Y ~ U(a,b) through location-scale
+      # transformation: Y = a + (b-a) X. see [GROESE11], section 3.1.2.2.
+      @min + @range * @rng.rand
     end
   end
 

data/lib/erv/version.rb

@@ -1,3 +1,3 @@
 module ERV
-  VERSION = '0.0.2'
+  VERSION = '0.2.0'
 end

data/lib/erv.rb

@@ -1,10 +1,2 @@
-# Load Apache Commons Math 3.3 jar if we're using JRuby
-if RUBY_PLATFORM =~ /java/
-  JARS_DIR = File.expand_path(File.join(File.dirname(__FILE__), '..', 'jars'))
-  Dir["#{JARS_DIR}/*.jar"].each do |jar|
-    $CLASSPATH << jar unless $CLASSPATH.include?(jar)
-  end
-end
-
 require 'erv/version'
 require 'erv/random_variable'

data/{spec/erv/constant_distribution_spec.rb → test/erv/constant_distribution_test.rb}

@@ -1,4 +1,4 @@
-require 'spec_helper'
+require 'test_helper'
 
 require 'erv/constant_distribution'
 
@@ -7,21 +7,19 @@ describe ERV::ConstantDistribution do
   it 'should require a reference value' do
     lambda do
       ERV::ConstantDistribution.new
-    end.should raise_error
+    end.must_raise ArgumentError
   end
 
   context 'reference value' do
 
     it 'should be accepted at initialization time' do
-      lambda do
-        ERV::ConstantDistribution.new(value: 10)
-      end.should_not raise_error
+      ERV::ConstantDistribution.new(value: 10)
     end
 
     it 'should match the value returned by sampling' do
       val = rand(100)
       crv = ERV::ConstantDistribution.new(value: val)
-      crv.sample.should == val
+      crv.sample.must_equal val
     end
 
   end

data/test/erv/distribution_test.rb

@@ -0,0 +1,17 @@
+require 'test_helper'
+
+require 'erv/distribution'
+
+describe ERV::Distribution do
+
+  context 'when explicitly given an RNG' do
+
+    it 'should use the given RNG' do
+      rng = Random.new
+      d = ERV::Distribution.new(rng: rng)
+      d.instance_variable_get(:@rng).must_equal rng
+    end
+
+  end
+
+end

data/test/erv/mixture_distribution_test.rb

@@ -0,0 +1,55 @@
+require 'test_helper'
+
+require 'erv/mixture_distribution'
+
+describe ERV::MixtureDistribution do
+  it 'should require at least two distributions' do
+    lambda do
+      ERV::MixtureDistribution.new([ { distribution: :exponential, rate: 1.0, weight: 0.5 } ])
+    end.must_raise ArgumentError
+  end
+
+  it 'should keep track of distribution weights (for normalization)' do
+    # create a mixture distribution with unnormalized weights
+    uw_md = ERV::MixtureDistribution.new([ { distribution: :exponential, rate: 1.0, weight: 100.0 },
+                                           { distribution: :exponential, rate: 2.0, weight: 200.0 },
+                                           { distribution: :exponential, rate: 3.0, weight: 300.0 } ])
+    uw_md.instance_variable_get("@weight_sum").must_equal 600.0
+  end
+
+  let :md do
+    ERV::MixtureDistribution.new([ { distribution: :exponential, rate: 1.0, weight: 0.3 },
+                                   { distribution: :exponential, rate: 2.0, weight: 0.2 },
+                                   { distribution: :exponential, rate: 3.0, weight: 0.5 } ])
+  end
+
+  context 'sampling' do
+
+    it 'should allow sampling from the mixture' do
+      md.sample
+    end
+
+  end
+
+  context 'moments' do
+
+    let :expected_mean do
+      0.3 * 1/1.0 + 0.2 * 1/2.0 + 0.5 * 1/3.0
+    end
+
+    let :expected_variance do
+      0.3 * (1/1.0 - expected_mean) ** 2 + (1/1.0) ** 2 +
+      0.2 * (1/2.0 - expected_mean) ** 2 + (1/2.0) ** 2 +
+      0.5 * (1/3.0 - expected_mean) ** 2 + (1/3.0) ** 2
+    end
+
+    it 'should correctly calculate the mean of the mixture' do
+      md.mean.must_equal expected_mean
+    end
+
+    it 'should correctly calculate the variance of the mixture' do
+      md.mean.must_equal expected_mean
+    end
+
+  end
+end

data/test/erv/random_variable_test.rb

@@ -0,0 +1,30 @@
+require 'test_helper'
+
+require 'erv/random_variable'
+
+describe ERV::SequentialRandomVariable do
+
+  it 'should require the :first_value parameter' do
+    lambda do
+      ERV::SequentialRandomVariable.new
+    end.must_raise ArgumentError
+  end
+
+  it 'should consider starting value' do
+    first = 1.0
+    srv = ERV::SequentialRandomVariable.new(first_value: first, distribution: :exponential, rate: 2.0)
+    srv.next.must_be :>, first
+  end
+
+  it 'should consider previous sample' do
+    first = 1.0
+    srv = ERV::SequentialRandomVariable.new(first_value: first, distribution: :exponential, rate: 2.0)
+    previous_sample = srv.next
+    10.times do
+      new_sample = srv.next
+      new_sample.must_be :>, previous_sample
+      previous_sample = new_sample
+    end
+  end
+
+end

data/test/test_helper.rb

@@ -0,0 +1,6 @@
+require 'minitest/autorun'
+require 'minitest/spec'
+require 'minitest-spec-context'
+
+require 'minitest/reporters'
+Minitest::Reporters.use!

metadata

@@ -1,72 +1,86 @@
 --- !ruby/object:Gem::Specification
 name: erv
 version: !ruby/object:Gem::Version
-  version: 0.0.2
+  version: 0.2.0
 platform: ruby
 authors:
 - Mauro Tortonesi
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-02-04 00:00:00.000000000 Z
+date: 2017-01-26 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ~>
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.3'
+        version: '1.14'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ~>
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '1.3'
+        version: '1.14'
 - !ruby/object:Gem::Dependency
   name: rake
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - '>='
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
+        version: '12.0'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - '>='
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
+        version: '12.0'
 - !ruby/object:Gem::Dependency
-  name: rspec
+  name: minitest
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - '>='
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
+        version: '5.10'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - '>='
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
+        version: '5.10'
 - !ruby/object:Gem::Dependency
-  name: gsl-nmatrix
+  name: minitest-reporters
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - '>='
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
-  type: :runtime
+        version: '1.1'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.1'
+- !ruby/object:Gem::Dependency
+  name: minitest-spec-context
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: 0.0.3
+  type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - '>='
+    - - "~>"
       - !ruby/object:Gem::Version
-        version: '0'
-description: erv-0.0.2
+        version: 0.0.3
+description: erv-0.2.0
 email:
 - mauro.tortonesi@unife.it
 executables: []
@@ -77,6 +91,7 @@ files:
 - LICENSE
 - README.md
 - Rakefile
+- TODO
 - erv.gemspec
 - lib/erv.rb
 - lib/erv/constant_distribution.rb
@@ -87,13 +102,16 @@ files:
 - lib/erv/gaussian_distribution.rb
 - lib/erv/general_pareto_distribution.rb
 - lib/erv/geometric_distribution.rb
+- lib/erv/mixture_distribution.rb
 - lib/erv/random_variable.rb
-- lib/erv/rng.rb
 - lib/erv/support/try.rb
 - lib/erv/uniform_distribution.rb
 - lib/erv/version.rb
-- spec/erv/constant_distribution_spec.rb
-- spec/spec_helper.rb
+- test/erv/constant_distribution_test.rb
+- test/erv/distribution_test.rb
+- test/erv/mixture_distribution_test.rb
+- test/erv/random_variable_test.rb
+- test/test_helper.rb
 homepage: https://github.com/mtortonesi/ruby-erv
 licenses:
 - MIT
@@ -104,20 +122,23 @@ require_paths:
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
-  - - '>='
+  - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
-  - - '>='
+  - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.1.11
+rubygems_version: 2.6.8
 signing_key: 
 specification_version: 4
 summary: Easy/elegant Ruby library providing support for random variable generation
 test_files:
-- spec/erv/constant_distribution_spec.rb
-- spec/spec_helper.rb
+- test/erv/constant_distribution_test.rb
+- test/erv/distribution_test.rb
+- test/erv/mixture_distribution_test.rb
+- test/erv/random_variable_test.rb
+- test/test_helper.rb

data/lib/erv/rng.rb

@@ -1,29 +0,0 @@
-if RUBY_PLATFORM == 'java'
-  require 'java'
-  java_import org.apache.commons.math3.random.MersenneTwister
-else
-  require 'gsl'
-end
-
-
-module ERV
-
-  class RNG
-    def self.make_rng(seed = nil)
-      # if not explicitly provided, seed is taken from the (lower quality)
-      # pseudo-random Kernel::rand generator
-      if RUBY_PLATFORM == 'java'
-        # this is a somewhat ugly workaround in order to avoid ambibuities in
-        # the constructor method that will be called (we basically explicit the
-        # signature of the constructor that we want to use)
-        constructor = org.apache.commons.math3.random.MersenneTwister.java_class.constructor(Java::long)
-        rng = constructor.new_instance(seed || Kernel::rand(2**31 - 1))
-      else
-        rng = seed ?
-          GSL::Rng.alloc(GSL::Rng::MT19937, seed) :
-          GSL::Rng.alloc(GSL::Rng::MT19937, Kernel::rand(2**31 - 1))
-      end
-    end
-  end
-
-end

data/spec/spec_helper.rb

@@ -1,20 +0,0 @@
-# This file was generated by the `rspec --init` command. Conventionally, all
-# specs live under a `spec` directory, which RSpec adds to the `$LOAD_PATH`.
-# Require this file using `require "spec_helper"` to ensure that it is only
-# loaded once.
-
-# Required to setup classpath for Apache Commons Math 3.3-SNAPSHOT in JRuby
-require 'erv'
-
-# See http://rubydoc.info/gems/rspec-core/RSpec/Core/Configuration
-RSpec.configure do |config|
-  config.treat_symbols_as_metadata_keys_with_true_values = true
-  config.run_all_when_everything_filtered = true
-  config.filter_run :focus
-
-  # Run specs in random order to surface order dependencies. If you find an
-  # order dependency and want to debug it, you can fix the order by providing
-  # the seed, which is printed after each run.
-  #     --seed 1234
-  config.order = 'random'
-end